SONICS: A new Peruvian hydrological dataset of simulated daily streamflow for flood monitoring and forecasting

Increasing hydrological risks and water use puts pressure on water resources and highlights the importance of a systematic hydrological analysis and modeling at national scale in gauged and ungauged catchments. This paper aims to develop a national hydrological model using physiographic and climatic characteristics to compute dissimilarity indices to pair donor and receptor sub-basins for the entire Peru. Therefore, we use the gridded hydrometeorological PISCO dataset (0.1º x 0.1º) to drive a conceptual rainfall-runoff (ARNO/VIC) model, which serves as an input for a river-routing (RAPID) model in thousands of river reaches. We identify 122 similarity-based hydrological zones across the country to run the hybrid model (ARNO/VIC+RAPID) with previously calibrated parameters. National daily streamflow simulations show good performance (KGE ≥ 0.75, NSE_sqrt ≥ 0.65, MARE ≤ 2, and -25% ≤ PBIAS ≤ 25%) for catchments located at the Pacific coast and the Andes-Amazon transition. Finally, a new hydrological dataset of daily flow series for entire Peru is presented, including a temporal coverage from 1 January 1981 to 31 March 2020. This new product represents an important contribution for water resource modeling including future risk scenario simulations in often poorly gauged catchments in Peru.

Keywords: Peru; PISCO; hydrological regionalization; large-scale modeling; national streamflow simulation.